Voltage tripler with PWM signals

[ronsimpson]

What's exactly the difference between using that or a regular voltage doubler with capacitors and diodes?

No difference. It just does not use any recesses from your computer.

Question: If we make a "power supply" to get 10 to 15 volts, what high frequency tone can you produce?
50khz to 150khz square wave, 50%/50%?????
25%/75%????
Some signal that you turn on and leave on all the time.

 

[Gabriel Sá Pinto]

No. The On-resistance of the MosFet and the 0.66Ω load resistor form a voltage divider. If the Rds(on) is 0.600mΩ, what is the voltage across the load? What is the current through the load? You dont need LTSpice to answer those questions. To get most of the 3.3V supply voltage across the load, what would the Rds(on) have to be compared to 0.66Ω?

I know the Rds has to be lower, the thing is I'm not being able to find a MOSFET that satisfies all the requirements.

What is the load? What is it used for? How are you going to create a varying control voltage for the gate? Is the NMOS transistor used as an on-off switch, or are you trying to use it as a linear amplifier, where the current in the load is proportional to a varying voltage created by the MSR?

I've said it before, I'm testing PSU rails, like 9 of them, all have different Voltages and current thresholds, but the highest current I'll be putting through the load will be 2.5A for a 3.3V supply, so I'm doubling those specs for safety precautions, I'm assuming a load that can handle 5A or for this case close to 15W power. If you need a better overview at what I'm doing check out this link:
https://www.instructables.com/id/Arduino-Programmable-Constant-Current-Power-Resist/

It's the same thing except I won't be using such a high power application, and I won't be going as fancy as including fans and encoders, I just want to run the tests from a serial interface, in that interface I'll be turning the IPSU rails on and off digitally and for each one I'll increase the current through the load until they reach their thresholds, so I'm guessing your second option, "using it as a linear amplifier" is what I'm going for.

Who says you have to have a Spice model to do your design?

Well I'm not that experienced with LTSPICE, been using it for like 2 months and only for some tests on previously assembled designs. I just wanted to have the model so I could be sure the MOSFET would respond well with my design.

To be honest I'm getting a little bit overly confused, and I'm guessing this is simpler than what it looks like, because I'm reading so much of these type of applications I really don't know what to do anymore.
I was presented with 5 main solutions:
- Double or triple the DC voltage of that PWM signal so I could use an easy to find power MOSFET
- Use the solution proposed by ronsimpson which I designed 2 posts above
- Use a level translator IC
- Find a MOSFET capable of being directly driven by a rectified PWM signal at a maximum of 3.3 voltage.
- I've also seen this link in which they use a Gate driver, but I'm still trying to understand if this applies to my project or not (https://www.microchip.com/forums/m855528.aspx)

So that's basically it I guess,I don't think there's any detail left. I just need to understand what to do next.

 

[Mosaic]

I had to handle something like this a while back. I believe you have a DC3.3V supply. The circuit generates its own 'PWM' for boosting the voltage. You can supply your own if you have a PWM source of reasonable current capability (up to 130 mA spikes on boot settling to 70mA)
Here's a circuit that will do 2 things with a load of about 1k which is ok for FET gates as the output cap will provide the current transient for the gate charge.
1) provide an isolated gate drive for the FET of around Vpeak -3.3V with SCOM as the 'ground'
2) Provide a Vpeak gate drive of about 12V if you use the common ground.

Edit: If you want to adjust the voltage out, replace the 1000Ω ouput resistor with a 1K pot. Lower the resistance to lower the output voltage .

This liberates you to use cost effective low RDs On NFETs with the 12V rail driving the gate. If you remove the burden 1K resistor you will probably need a 12 to 18V zener to clamp that charge pump gate voltage from rising too high.

 

[schmitt trigger]

Very clever Mosaic!

I love discretes-only circuits.

 

[audioguru]

You showed a simulation with an awful LM324 opamp. Its datasheet shows that its output cannot go higher than about 1.2V less than its positive supply and it performs poorly above 2kHz.

 

[Mosaic]

You showed a simulation with an awful LM324 opamp. Its datasheet shows that its output cannot go higher than about 1.2V less than its positive supply and it performs poorly above 2kHz.

say what?

 

[Pedro Ash]

Using a 12V supply would really going to work.

 

[Gabriel Sá Pinto]

You showed a simulation with an awful LM324 opamp. Its datasheet shows that its output cannot go higher than about 1.2V less than its positive supply and it performs poorly above 2kHz.

People have done more complex designs with LM324 amps with no side effects, just check this one: https://www.instructables.com/id/Arduino-Programmable-Constant-Current-Power-Resist/
I may be overwhelmed by this project but in the middle of all components the one I know won't bring me any issues is the LM324.

I had to handle something like this a while back. I believe you have a DC3.3V supply. The circuit generates its own 'PWM' for boosting the voltage. You can supply your own if you have a PWM source of reasonable current capability (up to 130 mA spikes on boot settling to 70mA)
Here's a circuit that will do 2 things with a load of about 1k which is ok for FET gates as the output cap will provide the current transient for the gate charge.
1) provide an isolated gate drive for the FET of around Vpeak -3.3V with SCOM as the 'ground'
2) Provide a Vpeak gate drive of about 12V if you use the common ground.

Edit: If you want to adjust the voltage out, replace the 1000Ω ouput resistor with a 1K pot. Lower the resistance to lower the output voltage .

This liberates you to use cost effective low RDs On NFETs with the 12V rail driving the gate. If you remove the burden 1K resistor you will probably need a 12 to 18V zener to clamp that charge pump gate voltage from rising too high.
View attachment 107042

This seems interesting but I'm guessing it'll be too complicated for me, I was aiming for something simpler. Also I was aiming for something controlled with a micro controller not with a pot in the load.

Using a 12V supply would really going to work.

It would to an extent, it's the same case as the voltage tripler, I can boost the voltage from the PWM all I want, but if the MOSFET hasn't got the right specs I'm screwed.
Bottom line:

• If I use one of the MOSFETS on my initial designs, despite the unappropriated specs, as Mike mentioned, I could get it to work, with a voltage tripler, or an external supply with another FET to translate the voltage level, but the MOSFET I want to drive won't turn on all the way because of the RDS, I still could get to work by slightly changing the load since I won't be needing more than 2.6 A to test the PSU rails anyway, but I feel I'm building something in "cardboard" with no solid foundation that way.
• Another solution was to find a MOSFET with a lower RDS and Vth with enough power to sustain this application, so it would be turned on all the way to max current by 3.3 gate voltage. I've been there done that, but all the MOSFETs I've found that fit in those specs will completely turn on (from 0 to 5A) only with a couple mV of difference in the gate, which is inappropriate for my application as well, since I want to gradually increase the current going though the power supply to test, not all at once.

I'm once again stuck. I hope I'm not being ungrateful with all the possible solutions that were presented to me, I'm much appreciated by all your efforts in trying to help me.
Also I'm not very experienced in the field of power electronics so I still lack some understanding in some things, I hope you give some slack if I don't understand something right away.

Thank you all once again

 

[MikeMl]

Ignore what audioguru says about the LM324. The Instructable article is using it in a mode where its 0.5V/us slew rate doesn't matter. Neither does that it isn't rail-to-rail out.

The LM324 has a huge advantage in this application: that is that it's common-mode input range can go 0.5V below Vee!

 

[Mosaic]

This seems interesting but I'm guessing it'll be too complicated for me, I was aiming for something simpler. Also I was aiming for something controlled with a micro controller not with a pot in the load.

Ok, here is a PWM driven option that drives voltage up via PWM DC%.

 

[ronsimpson]

here is a PWM

The output voltage is very dependent on load. You could replace R4 with a Zener.

 

[Mosaic]

Or he could use the uC ADC to sense the voltage, perhaps by splitting R4 into a voltage divider to bring the voltage into the ADC range. Then feedback with the PWM DC% for a variable supply under uC control. From the time constant of the circuit a uC should have ample time to execute control.

 

[ronsimpson]

Or he could use the uC ADC to sense the voltage

A flyback power supply can produce a very high voltage. (high voltage in a old TV)
I do not trust software. Production software is one thing but trouble shooting software problems is very dangerous when it also watches a PWM.
I would put a Zener or some thing to limit the voltage if the software goes nuts.

 

[Gabriel Sá Pinto]

Sorry for no response for over a week, this project has been on hold for a while.
A colleague of mine suggested a MOSFET Gate Driver, we're really aiming for simplicity in the design (keep in mind I'll 9 of these circuits on the same board).
The thing is I don't understand how it would work since a Gate Driver can output a "drivable" voltage and current for a common MOSFET but it doesn't let me control how much turned on the MOSFET is... in simpler words what I mean is that the GATE driver outputs 10v for example, but that voltage stays the same no matter how much I change the duty cycle of the PWM. Or am I just looking at this thing all wrong?

Thank you.

 

[alec_t]

If you have a 10V gate drive signal which is PWM-ed, the duty cycle of the PWM determines the average 'on' time of the FET and hence the average load current or voltage. The FET is fully on or fully off. This is preferable to having it partially on (which causes it to waste power and get hot).

 

[Gabriel Sá Pinto]

If you have a 10V gate drive signal which is PWM-ed, the duty cycle of the PWM determines the average 'on' time of the FET and hence the average load current or voltage. The FET is fully on or fully off. This is preferable to having it partially on (which causes it to waste power and get hot).

I understand that, but how does this allow me to control the current going though the MOSFET? Because that's what I need, I know it'll waste power but that's why everyone uses a heatsink in this kind of applications. The only way I see of controlling current is leaving the MOSFET partially on, but gate drivers don't seem to let me do that.

 

[audioguru]

Pulse Width Modulation: the switching device is fully turned on with hardly any voltage across it so it does not consume much power, then it is fully turned off with no current in it so it consumes no power. The switching device needs a heatsink only when the on-current is very high.
Wide pulses: then the load gets high average power.
Narrow pulses: then the load gets low average power.

The PWM pulse widths adjust the average current then you do not need to control the current in the switching device.

 

[audioguru]

In post #22 you linked to An Instructable that is a linear constant current load, it does not use PWM. Its Mosfet is linear so it gets extremely hot by wasting power. PWM is used with a heater or with an LED where you do not notice its power switching on and off quickly.

 

[Gabriel Sá Pinto]

In post #22 you linked to An Instructable that is a linear constant current load, it does not use PWM. Its Mosfet is linear so it gets extremely hot by wasting power. PWM is used with a heater or with an LED where you do not notice its power switching on and off quickly.

Yeah but a linear constant current load is what I want, I was planning to use PWM and then rectify it with an RC filter, my goal is NOT to drive directly by PWM.
I've done it before with a VCO but back then I did on Arduino and it was way simpler.
Now the principle is the same but instead of feeding a VCO I want to feed a MOSFET's gate to control the current going through, like a variable resistor. Varying the duty cycle I should get different voltages at the mosfet's gate, therefore allowing more or less current though the mosfet's drain, am I right?
Is it so different I'm not getting it? How would you do it?

 

[alec_t]

I should get different voltages at the mosfet's gate, therefore allowing more or less current though the mosfet's drain, am I right?

Yes BUT the current will also depend on the load and the particular FET's characteristics, so won't be very predictable.